Multimodal Clinical Data Integration and Artificial Intelligence Modeling for Predicting Complications Following Pediatric Transcatheter Closure of Perimembranous Ventricular Septal Defect

What this Trial Is About

Researchers are developing and validating an artificial intelligence (AI) model to predict treatment-related complications in children with perimembranous ventricular septal defect (pmVSD) who have undergone transcatheter device closure. This observational study aims to determine if combining information from demographics, lab results, health records, echocardiography, chest X-rays, and electrocardiograms can accurately assess individual risk of complications within 30 days after the procedure. The study uses retrospective clinical data from pediatric patients treated at a single center. The AI model will be trained using deep learning techniques to analyze various types of clinical data, including text and images, collected from routine care records. The study does not involve any new treatments or interventions but focuses on data integration and AI prediction methods. The primary outcome is the occurrence of procedure-related complications within 30 days after the transcatheter closure. Participants are children aged 18 years or younger who had a confirmed diagnosis of pmVSD and underwent transcatheter device closure at the study center. Researchers will review medical records to gather necessary clinical and imaging data. The study evaluates the AI model's performance in predicting complications using metrics such as precision-recall curves, sensitivity, and predictive values, all within 30 days after the procedure. The study is sponsored by Xinhua Hospital, Shanghai Jiao Tong University School of Medicine, and is planned to end in June 2026.

A Multimodal AI Prediction Model for Complications After Transcatheter Closure of Perimembranous VSD in Children